This repository contains all of the code required to run the protein-sol abpred machine learning models trained on the Jain 2017 dataset of 137 clinical stage therapeutic antibodies sequences and performance on 12 biophysical characterisation platforms.

The paper describing this work has been submitted for consideration and the preprint is available here.

This code can run predictions on multiple proteins.

There is a docker image preconfigured with all necessary software and dependencies to run this software here.

For this project, we have supplied a list of yaml configuration file (abpred_env.yml) that contains a list of all packages available for installation from conda. If you already have conda installed you can run the install.sh script that sets up the conda environment and also installs the required bestNormalize package for the mathematical transformations.

The software is designed to work with the following languages and versions but may work with other versions.

bash, perl 5, R 3.5.1, python3.6

For these languages we also require the following R packages

randomforest, caret, glmnet, kernlab, bestNormalize

and pandas and its dependencies for python.

The run_preds.sh shell script is a simple pipeline for running the required code. It requires two arguments to run, the fasta file and an optional identification name. For example

run_preds.sh test.fasta job1

In your current directory, the script will return a file for each of the biophysical characterisation platforms as well as a file containing sequence features for each of the input Fv.